Editing Overlays in QC report - dwi threshold used for extraction & number slices shown for MELODIC

[dsodums]

I was wondering if its simple to change the QC code for the dwi extraction default threshold used when displaying the B0 image underlay as when using the Rockland dataset with dwi bval=1500 the image is really faint.

Also, is it possible to change how many slices are shown for the MELODIC report?

Thanks,
Devin

[justinliuyuwang]

We are still in the process of fleshing out the Modifying QC Generation page, but you can find some details for modifying the B0 underlay there.

Currently, calls to lightbox.sh from tvb-ukbb/tvb_bb_QC/tvb_bb_QC.sh generate the dwi extraction B0 image overlays. The -b and -c options correspond to the brightness and contrast of the underlay image in the final overlay image. Please note that there are two calls to lightbox.sh for the extraction analysis (one unmasked and one masked). You will just need to fiddle with the -b and -c values for those two calls to lightbox.sh until you find a setting that works for your dataset.

Unmasked

#dMRI extraction unmasked
		$BB_BIN_DIR/tvb_bb_QC/lightbox.sh -b 62.718401747457314 -c 96.08735171986915 /dMRI/dMRI/data_B0.nii.gz $dirSubject dMRI_extraction_unmasked

Masked

	#dMRI extraction masked
		$BB_BIN_DIR/tvb_bb_QC/lightbox.sh -o /dMRI/dMRI/nodif_brain_mask.nii.gz -b 62.718401747457314 -c 96.08735171986915 /dMRI/dMRI/data_B0.nii.gz $dirSubject dMRI_extraction_masked

Unfortunately, the images shown in the MELODIC and EDDY_QUAD reports are generated by their respective programs. The workaround would be to add calls to tvb-ukbb/tvb_bb_QC/lightbox.sh and/or tvb-ukbb/tvb_bb_QC/edges.sh into tvb-ukbb/tvb_bb_QC/tvb_bb_QC.sh so that the QC sub-pipeline will generate the desired slices. These new images would not appear on the QC Report, however; they would only appear as png files in the images directory of the subject's QC folder. We are working on a way for users to have their new QC images included into the html report.

[dsodums]

Thank you! Apologies that I didn't see the QC modifying page before, that's very helpful and should give me what I need to play around with the threshold values for display.

Understood regarding the MELODIC outputs, thanks again for the help :)

[justinliuyuwang]

No problem and no worries! Those pages are there for us to point to when users have questions :)

[dsodums]

I'm actually a bit confused, I went to edit those two lines in tvb-ukbb/tvb_bb_QC/tvb_bb_QC.sh but they look very different in my version of the file than what you had pasted:

#dMRI extraction unmasked
$BB_BIN_DIR/tvb_bb_QC/lightbox.sh /dMRI/dMRI/data_B0.nii.gz $dirSubject dMRI_extraction_unmasked

            #$BB_BIN_DIR/tvb_bb_QC/image_gen_link.sh $dirSubject "uw_under" "../../dMRI/dMRI/data_B0.nii.gz" "NA22" "NA22_link" "NA23" "NA23_link" 0


    #dMRI extraction masked
            $BB_BIN_DIR/tvb_bb_QC/lightbox.sh -o /dMRI/dMRI/nodif_brain_mask.nii.gz /dMRI/dMRI/data_B0.nii.gz $dirSubject dMRI_extraction_masked


            $BB_BIN_DIR/tvb_bb_QC/image_gen_link.sh $dirSubject "dex_under" "../../dMRI/dMRI/data_B0.nii.gz" "dex_over1" "../../dMRI/dMRI/nodif_brain_mask.nii.gz" "NA24" "NA24_link" 0

Am I missing something?

[justinliuyuwang]

Sorry, the two lines to be changed are in lines 190 (unmasked) and line 197 (masked). The main branch does not have brightness and contrast options for this analysis. When you add these options, they should be placed right after the $BB_BIN_DIR/tvb_bb_QC/lightbox.sh in the call, like what I had pasted in my previous post.

[dsodums]

Oh perfect, thank you so much!

I'm hoping to test just the QC stage using an interactive session on Compute Canada rather than submitting the entire pipeline. I have that session running but what would be the easiest way to test just this stage? I've played around with pasting just that part of the code but think I'm missing something so any pointers would be greatly appreciated.

[justinliuyuwang]

In tvb-ukbb/bb_pipeline_tools/bb_pipeline.py I would comment out calls to other subpipes, i.e. remove the following lines (you may have to look around the file for them, they are not on consecutive lines):

bb_pipeline_struct(subject, runTopup, fileConfig)
bb_pipeline_func(subject, fileConfig)
bb_pipeline_diff(subject, fileConfig)
bb_IDP(
        subject, fileConfig
    )

You will then be able to run the pipeline as normal and it will only run the QC subpipeline.

[dsodums]

Thank you so much!!